Guess the 2019 Intel desktop product!

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

What will Intel release for mainstream desktop in 2019?

  • Icelake (as originally intended with 8 CPU cores/48 EU max)

    Votes: 21 26.6%
  • Icelake (with core counts/GPU reduced)

    Votes: 3 3.8%
  • "Champagne Lake" (14++ Coffee Lake with some minor improvements)

    Votes: 35 44.3%
  • Rushed out Tiger Lake/ Core EMIB

    Votes: 0 0.0%
  • Moar Coffee Lake!

    Votes: 13 16.5%
  • Nothing!!

    Votes: 5 6.3%
  • Other

    Votes: 2 2.5%

  • Total voters
    79
  • Poll closed .

CatMerc

Golden Member
Jul 16, 2016
1,114
1,149
136
From what insiders can tell, 10nm is so broken even a 2019 product is in question. David Schor being the main one.

Therefore I'd say the safe bet is another form of Skylake based design, with Intel likely following the way Coffee Lake launched. IE put out K SKU's with 8c i7 and 6/12 i5 in August 2018, and the finish up the lineup around Q1 2019.

They haven't ported Icelake to 14nm, so the question is if they can get 10nm working in time for a Q4 2019 limited launch.
 
  • Like
Reactions: exquisitechar

turtile

Senior member
Aug 19, 2014
614
294
136
Whiskey Lake is for mobile not desktop.

The rumor says it's for mobile but it's not stated that way by Intel:

We continue to make progress on our 10-nanometer process. We are shipping in low volume and yields are improving, but the rate of improvement is slower than we anticipated. As a result, volume production is moving from the second half of 2018 into 2019. We understand the yield issues and have defined improvements for them, but they will take time to implement and qualify. We have leadership products on the roadmap that continue to take advantage of 14-nanometer, with Whiskey Lake for clients and Cascade Lake for the data center coming later this year.

OEMs also want a new product yearly so this seems to make sense. They have no choice but to fill in another product. It also makes more sense to start 10nm on a mobile product due to higher yield due to size.
 

ksec

Senior member
Mar 5, 2010
420
117
116
If 10nm is not ready for desktop HVM, there isn't much else they can do. I doubt they will try to get TSMC to fab them chips. ;)

It isn't so much about Desktop HVM, it is even if they can get HVM intel will likely prioritise DC over Desktop. DC is where the growth, money and profits are, and they can't stand by and let Qualcomm eat into it, since Centriq is going to get IPC and 7nm product there as well. All capacity will be DC first, Mobile second, and Desktop last.

7nm AMD GPU will be all TSMC, which means there is more capacity form GF's on 7nm, and their only job is to get 7nm up asap. The only way Intel will lower its margin is when AMD give them some pressure, but I assume AMD wants as much profits as possible rather then market share.
 

jpiniero

Lifer
Oct 1, 2010
14,509
5,159
136
It isn't so much about Desktop HVM, it is even if they can get HVM intel will likely prioritise DC over Desktop. DC is where the growth, money and profits are, and they can't stand by and let Qualcomm eat into it, since Centriq is going to get IPC and 7nm product there as well. All capacity will be DC first, Mobile second, and Desktop last.

For Intel to go HVM, I take it to mean yield would have to be good enough to do tiny mobile parts, not anything huge like a monolithic server die. That's where EMIB comes in but it's not entirely clear when Sapphire Rapids will be ready.
 

mikk

Diamond Member
May 15, 2012
4,111
2,105
136
The rumor says it's for mobile but it's not stated that way by Intel:


Doesn't matter, WHL is a mobile part, it's just a KBL-R refresh with 4 cores on 14nm++. For desktop there is CFL 8C coming.
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
By all accounts sacrificing margins is the very last thing Intel will do.

They really should. Public companies have the problem of eventually being driven by shareholder concerns, and they are all short term. There needs to be a company brave enough to say "we care about long term, so we'll sacrifice few % revenue/margin/profits". Companies like Intel are successful enough that they shouldn't worry about share prices so much.
 
  • Like
Reactions: Space Tyrant

PeterScott

Platinum Member
Jul 7, 2017
2,605
1,540
136
They really should. Public companies have the problem of eventually being driven by shareholder concerns, and they are all short term. There needs to be a company brave enough to say "we care about long term, so we'll sacrifice few % revenue/margin/profits". Companies like Intel are successful enough that they shouldn't worry about share prices so much.

Generally the most successful companies are the ones with the best margins. Those rich margins allow them to invest in more R&D, to drive future profits.

There is no reason for Intel to sacrifice margins unless they are showing a significant loss in sales, and really they haven't yet shown that.
 
  • Like
Reactions: ksec

ehume

Golden Member
Nov 6, 2009
1,511
73
91
Can we hope for enthusiast chips with no igpu wasted space?
Hey. I'm a regular user and I haven't bought a dGPU since Intel switched to iGPU's. AFAIAC, the Images I get on my 24" 1920x1080 from Intel's integrated GPU's are Good Enough(tm).
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
What does that have to do with sacrificing margins?

Absolute push for margins are why they happen to make otherwise decent products, but not price competitive. Margins are also why they were reluctant to enter the Smartphone market in the first place, and why excessive segmentation exist.
 

ehume

Golden Member
Nov 6, 2009
1,511
73
91
I just hope they give me some compelling reason to upgrade from my 4790k.
There is not, nor is there likely to be a product giving that compelling reason in this decade (which will end after 2020).
 

PeterScott

Platinum Member
Jul 7, 2017
2,605
1,540
136
Absolute push for margins are why they happen to make otherwise decent products, but not price competitive. Margins are also why they were reluctant to enter the Smartphone market in the first place, and why excessive segmentation exist.

You keep flipping back and forth. Not being price competitive only matters, if people stop buying your stuff because of it.

No company wants lower margins. Low margins are usually a sign you are producing commodity products, or that you are the underdog trying to take market share.

If you already have the biggest market share, there is nothing gained by lowering your margins.
 
  • Like
Reactions: ksec

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
No company wants lower margins.
If you already have the biggest market share, there is nothing gained by lowering your margins.

Sure there is. Intel decided to coast on the successes of X25-M, and Samsung quickly took the market away from them. They believed others won't advance. That coupled with their products being expensive. They should have waited until the SSD market grew, do everything to maintain #1 share, and from that point they could do whatever they want. That should be the approach for all new markets they enter.

And that's why Intel is still viewed as a PC company and nothing else. The advent of Smartphones and Tablets forced them to create small cores like Atom that sells for lot lower ASPs than they were used to. They segmented Netbooks from having higher resolution and greater memory, because they didn't want it to cannibalize Core. Once, they had the idea big Cores should be 10x the performance of small cores. Yea, they kept their dominant marketshare in PC. Competitors couldn't beat them in a traditional way, so they went a totally different way - Smartphones and Tablets.
 

PeterScott

Platinum Member
Jul 7, 2017
2,605
1,540
136
Sure there is. Intel decided to coast on the successes of X25-M, and Samsung quickly took the market away from them. They believed others won't advance. That coupled with their products being expensive. They should have waited until the SSD market grew, do everything to maintain #1 share, and from that point they could do whatever they want. That should be the approach for all new markets they enter.

Margins aren't the problem. They should have been using the margin, for more R&D to stay ahead.
 
  • Like
Reactions: whm1974

TheELF

Diamond Member
Dec 22, 2012
3,967
720
126
Most likely it will be a "14nm +++" ring bus 8 core along with a refresh of the 8700k and 8700. Intel will claim revolutionary new performance on offer with 33% more cores and they will show graphs that make it clear that intel is giving you more cores now than they ever have before.



Whether this marketing offensive + an 8 core actually works to take back marketshare from Ryzen... I have my doubts.
Why should Intel bother and do something like that?
The 6 cored 8700k, when highly clocked, already is faster then the 8 cored 2700x in a variety of software and not by a small margin but by like 30-40% faster.
Intel doesn't need to be faster in everything compared to a CPU with more cores, we all know this from the years of comparing to the FX line.
We already saw that AMD takes the same course with ryzen they took with the FX line,the 2700x has already been price cut from $500 that they asked for the 1800x to ~ $330, AMD already got to grip with reality,that they can't compete on hype alone and adjusted prices down.
Yes lisa told us that they won't be the cheap solution anymore but that's the only thing they can be.
 

exquisitechar

Senior member
Apr 18, 2017
655
862
136
Either more 14nm Skylake clone CPUs or late 2019 10nm Icelake. It's difficult to say.
Why should Intel bother and do something like that?
The 6 cored 8700k, when highly clocked, already is faster then the 8 cored 2700x in a variety of software and not by a small margin but by like 30-40% faster.
Intel doesn't need to be faster in everything compared to a CPU with more cores, we all know this from the years of comparing to the FX line.
We already saw that AMD takes the same course with ryzen they took with the FX line,the 2700x has already been price cut from $500 that they asked for the 1800x to ~ $330, AMD already got to grip with reality,that they can't compete on hype alone and adjusted prices down.
Yes lisa told us that they won't be the cheap solution anymore but that's the only thing they can be.
2700X wasn't "price cut", it's priced accordingly to the competitive landscape. The 1800X could be priced that high because it slaughtered everything Intel had on its mainstream desktop platform in terms of multithreaded performance, not "hype".
 

epsilon84

Golden Member
Aug 29, 2010
1,142
927
136
Why should Intel bother and do something like that?
The 6 cored 8700k, when highly clocked, already is faster then the 8 cored 2700x in a variety of software and not by a small margin but by like 30-40% faster.
Intel doesn't need to be faster in everything compared to a CPU with more cores, we all know this from the years of comparing to the FX line.
We already saw that AMD takes the same course with ryzen they took with the FX line,the 2700x has already been price cut from $500 that they asked for the 1800x to ~ $330, AMD already got to grip with reality,that they can't compete on hype alone and adjusted prices down.
Yes lisa told us that they won't be the cheap solution anymore but that's the only thing they can be.
Well Intel needs to release something new this year - since it won't be 10nm based, the logical conclusion would have to be an 8 core CFL to go along with the Z390 chipset launch.

That should give Intel a decent lead on the desktop, AMD has already shown it's cards for 2018, Intel has not, but it doesn't take a genius to figure out that 8C CFL should beat 8C Ryzen, given the higher IPC and clockspeeds from Intel.

The problem for Intel is succeeding the '9700K' with a 10nm chip that can perform well enough, from Intels own slide it can be seen that first gen 10nm actually has lower transistor performance compared to 14nm++, and that it will take a refresh on 10nm to reach 14nm++ levels of performance. This is problematic if 10nm keeps getting pushed back - Intel can't rely on 14nm indefinitely, especially with AMD on track for 7nm next year.
 

TheELF

Diamond Member
Dec 22, 2012
3,967
720
126
2700X wasn't "price cut", it's priced accordingly to the competitive landscape. The 1800X could be priced that high because it slaughtered everything Intel had on its mainstream desktop platform in terms of multithreaded performance, not "hype".
Ehh bottom line stays the same,even AMD realised that the 2700x is only worth as much as the 8700k at most,so there is no reason for intel to do anything other then they where doing for so many years...~10% better,split between IPC and clocks.
 

Midwayman

Diamond Member
Jan 28, 2000
5,723
325
126
Hey. I'm a regular user and I haven't bought a dGPU since Intel switched to iGPU's. AFAIAC, the Images I get on my 24" 1920x1080 from Intel's integrated GPU's are Good Enough(tm).

Sure. APUs will be a thing still I'm sure. Just better to have the cpu and gpu on different dies. Just mix and match parts later like intel is doing with AMD gpus if you want that. Otherwise you don't waste die space (and hence increase defect rates) on a gpu most people don't want. If I'm paying for the silicon Id rather have an extra core or two.
 

ehume

Golden Member
Nov 6, 2009
1,511
73
91
Sure. APUs will be a thing still I'm sure. Just better to have the cpu and gpu on different dies. Just mix and match parts later like intel is doing with AMD gpus if you want that. Otherwise you don't waste die space (and hence increase defect rates) on a gpu most people don't want. If I'm paying for the silicon Id rather have an extra core or two.
True, you can make that argument. But you have spent $150-1500 on GPU's, while I have spent $0.

And how much am I spending to include an iGPU? How much would I save to have a pure CPU? $10? $20? Whatever it is, I would spend more for a dGPU, a lot more.

And the concept of "most people." Most gamers is what you mean, I think. If you look at sale of dGPU cards I think it is a small fraction of the APU's that are sold. So "most people" are satisfied not to have to "mix and match parts" or to think about dGPU choices. That's what I meant when I said "regular user" in my first post.
 

Midwayman

Diamond Member
Jan 28, 2000
5,723
325
126
True, you can make that argument. But you have spent $150-1500 on GPU's, while I have spent $0.

And how much am I spending to include an iGPU? How much would I save to have a pure CPU? $10? $20? Whatever it is, I would spend more for a dGPU, a lot more.

And the concept of "most people." Most gamers is what you mean, I think. If you look at sale of dGPU cards I think it is a small fraction of the APU's that are sold. So "most people" are satisfied not to have to "mix and match parts" or to think about dGPU choices. That's what I meant when I said "regular user" in my first post.

I should clarify I'm thinking mostly of high end CPUs where even if you aren't a gamer people are using GPUs for compute or better video codecs and price isn't such an issue.

The issue with an igpu is on a 4 core chip the die are for it is like 1/3rd the chip. So just on a straight cost standpoint right now you are paying a substantial portion of your CPU cost as a iGPU. It also makes the die a lot larger. That's getting to be a bigger and bigger problem as process shrinks. The bigger the chip the worse your yield. The solution is what AMD is doing and using smaller dies and gluing things together. Undoubtedly that's what Keller will be working on now that he's at Intel. Making the Gpu on its own die and only adding it for people who want it in an APU package makes sense.

So, how much are you saving? Depends on the chip I guess. But the important thing is that its probably the path forward anyways. A apu will be cheaper than a chip and a low end dGPU certainly. Yield improvements with a design change probably means a apu would be cheaper than a cpu with on die iGPU.
 

PeterScott

Platinum Member
Jul 7, 2017
2,605
1,540
136
Why should Intel bother and do something like that?
The 6 cored 8700k, when highly clocked, already is faster then the 8 cored 2700x in a variety of software and not by a small margin but by like 30-40% faster.

To offer an expensive Halo part, that brings in more margin rich top end sales, and keeps the mind-share in the Intel camp. 8 core CL gives Intel the best mainstream socket CPU for everything. Why wouldn't they do that?

It is looking more certain that 8 Core Coffee Lake is coming in Q4 2018.
German IT Distributor Publishes AMD+Intel Roadmaps: Z490, Z390, 8-Core CFL in Q4

Which means it will be the top Intel mainstream socket CPU for most of 2019. So it will definitly win Intel late 2018 and early 2019, and it will probably even put up a good fight against 7nm Ryzen, though that may depend on how many cores AMD is bringing to mainstream and how fast 7nm GF clocks.